Canon R5 Comparison: Normal vs Hi-Res vs Topaz

[dmanthree]

I did a very quick test of the new hi-res feature included in the latest firmware, and here's the results.

The first is the full image, for reference. Then a 100% crop of the base image, the hi-res image, and an up-res'd image I did in Topaz AI, set to "Max." The results were a bit surprising.

 

[Bryan Conner]

Based on the images you presented here, the hi-res image looks awful to me and the Topaz-Max looks very good.

Are the hi-res and Topaz-Max images both displayed here at 100%?

Thanks for sharing this with us.

 

[koenkooi]

The stippling and banding in your IBIS HR image indicate there was movement during capture.

 

[dmanthree]

Based on the images you presented here, the hi-res image looks awful to me and the Topaz-Max looks very good.

Are the hi-res and Topaz-Max images both displayed here at 100%?

Thanks for sharing this with us.

Yes, all are at 100%. And I agree that the Topaz image looks a lot better than the OOC image. The lines on the OOC image look pretty bad. I'll give it some more tests soon, but for now, Topaz wins.

 

[dmanthree]

The stippling and banding in your IBIS HR image indicate there was movement during capture.

Camera was tripod mounted with IS turned off.

 

[koenkooi]

Camera was tripod mounted with IS turned off.

The HR mode disables lens IS regardless. You have movement artifacts in your image, so something moved.

In my tests I needed to let the camera settle for a few seconds after touching it, even when using a bluetooth remote. And that was with the camera on a macro rail flat on a 6cm thick oak tabletop.

 

[Bryan Conner]

The HR mode disables lens IS regardless. You have movement artifacts in your image, so something moved.

In my tests I needed to let the camera settle for a few seconds after touching it, even when using a bluetooth remote. And that was with the camera on a macro rail flat on a 6cm thick oak tabletop.

If that is what is needed to get acceptable results, then I would say the HR mode is flawed at best.

 

[dmanthree]

If that is what is needed to get acceptable results, then I would say the HR mode is flawed at best.

I've received some advice on the test, and need to shoot it again. I guess the mode is VERY sensitive to movement. I had the camera on a tripod, but did not use a remote trigger, so the test is flawed. I'll shoot again and post the results.

 

[koenkooi]

If that is what is needed to get acceptable results, then I would say the HR mode is flawed at best.

I completely agree with that. I did manage to get a decent focus stack using this mode, but it was very finicky, after getting everything stable, the LED lighting was causing artefacts

On top of that, Topaz Denoise takes 5.5 minutes per frame and Helicon Focus needs some more time as well, compared to 45MP RAWs.

 

[Bryan Conner]

I've received some advice on the test, and need to shoot it again. I guess the mode is VERY sensitive to movement. I had the camera on a tripod, but did not use a remote trigger, so the test is flawed. I'll shoot again and post the results.

I imagine that it works best on subjects without any movement at all. Wind blowing the leaves on a tree for example would probably result in sub-optimal results. Just like it would in focus stacking or stitched panoramas. Even with a high shutter speed the results might not be good.

It will be interesting to see your results and to hear your opinion. Thanks for sharing with us.

 

[Bryan Conner]

I completely agree with that. I did manage to get a decent focus stack using this mode, but it was very finicky, after getting everything stable, the LED lighting was causing artefacts

On top of that, Topaz Denoise takes 5.5 minutes per frame and Helicon Focus needs some more time as well, compared to 45MP RAWs.

From what I have read, older LED lights flicker and this could be what caused your artefacts. We don't normally see the flicker in these old lights, but a camera could capture this if the shutter is tripped at the exact moment as the flicker.

 

[MarshallG]

I've received some advice on the test, and need to shoot it again. I guess the mode is VERY sensitive to movement. I had the camera on a tripod, but did not use a remote trigger, so the test is flawed. I'll shoot again and post the results.

And not all tripods are created equal. Or there could be vibration from the road the tripod is on, etc.

I try to judge The Photograph.
What you’re doing is taking measurements and that’s pretty painstaking and you learn from it. But, the purpose is to ultimately learn what makes the ideal photograph. For 400mp, I assume this is for printing a billboard or a massive mural. I can’t say I know for certain, but I think it’s something like that.

Your learning process is helping a lot of people figure how how (of if) this new feature can be used. If a 100 pound tripod and exacting precision is what it takes, so be it.

 

[GrunRad]

After having installed 1.8.1., I took a few pictures out of my window. What I learned:

- you really need a very sturdy tripod
- you must use a remote trigger, or the 10 second self-timer
- the scene must be absolutely still
- air turbulence will make everything more that 100 feet away look like something created by a mad pointillistic painter

When all the stars align, there really is a lot of additional resolution - but I think that IBIS High Resultion in its current implementation addresses very specific use cases (like interior architecure or reproduction/scanning of scientific artefacts).

(left IBIS HiRes; right normal 45 MP from RAW, ACR with a bit of clarity added)
I used RF 50/1.2 at about 7.1 with manual focusing.
The red rectangle shows the area of the full image that is shown in the two crops.

 

[dmanthree]

After having installed 1.8.1., I took a few pictures out of my window. What I learned:

- you really need a very sturdy tripod
- you must use a remote trigger, or the 10 second self-timer
- the scene must be absolutely still
- air turbulence will make everything more that 100 feet away look like something created by a mad pointillistic painter

When all the stars align, there really is a lot of additional resolution - but I think that IBIS High Resultion in its current implementation addresses very specific use cases (like interior architecure or reproduction/scanning of scientific artefacts).

(left IBIS HiRes; right normal 45 MP from RAW, ACR with a bit of clarity added)
I used RF 50/1.2 at about 7.1 with manual focusing.
The red rectangle shows the area of the full image that is shown in the two crops.


View attachment 14320

Can you upload a shot that been up-res'd in something like Topaz for comparison? (*if* you have something like that...)

 

[RDM5546]

This Northrup's comparison of the Canon R5 feature with the Sony A7R5 was interesting to watch:

Canon R5 FREE 400 MEGAPIXELS beat Sony a1!!

Go to http://squarespace.com/Tony & save 10% off your first website or domain with code “Tony"Buy the Canon R5 at http://SDP.io/R5 and download the FREE firm...

www.youtube.com

 

[GrunRad]

Can you upload a shot that been up-res'd in something like Topaz for comparison? (*if* you have something like that...)

Of course!
Here are two (approximately) 4000x3000 central crops (the full image jpgs were too large for uploading) of the same scene.
First one is from the IBIS HiRes JPG, second one is from a 45 MB CR3, processed with DPP with camera standard, run through Topaz GigaPixel (3x).